29 May 1998 Kevin Karplus

T58 is UNG_ECOLI in Swissprot and is homologous to 1akz, 1lauE, and
1udiE (FSSP representative 1akz).  Note: 1udiE is the same as 1udg and
1udh.  No other PDB chain scores at all well, which is good, since
there are no other significant stuctural homologs in 1akz.fssp.

Since 1akz scores the best with the target model, and is the fssp
representative, it should probably be used as the structure to model from.

30 May 1998

1akz also scores best of the template models, and is the closest PDB
sequence in the evolutionary tree made from the t98_6 alignment.
I'll build the fssp-t98 and constr-t98 alignments for 1akz.


16 July 1998

Redid scoring using t58.t98_2.  Now 1akz gets -416.33 summing both
ways, and -210.3 with just the target model (top for both).
The tree t58.t98_2.tree has one subfamily with t58 and 1akz in it.

I selected out this subfamily as t58.close.a2m

Actually, there isn't much point to all this fussing---the alignment
is essentially gapless.  There is a one-residue insert in the middle,
the beginning may need up to 5 residues trimmed off, and the end may
need up to 3 residues trimmed off.


27 July 1998 Kevin Karplus

Moved the one-residue insert to the end of the helix, increasing the
residue identity.  Result is t58-1akz-hand.a2m.